The method
Read the record.
Test the explanation.
Project records research starts with a question, not a verdict. We connect documents, dates, decisions and payments so the next step rests on something another person can check.
Experience sets the question.
Technology does the sorting.
Emails, drawings, minutes, invoices and court filings rarely tell the same story in the same order. We identify the decision that matters, agree the record boundary and separate what is recorded from what still needs to be established.
Software helps index, compare and cross-reference material. AI can assist with organization and drafts. Neither replaces the source, resolves a missing approval or turns an allegation into a finding.
Keep the source attached.
Every material statement should lead back to an identifiable record. We preserve the document title or stable ID, author or issuing system, date, revision, status and custodian. A draft drawing, an issued revision and a signed instruction are different records, even when their wording overlaps.
We note missing pages, uncertain authenticity, extraction errors and limits on permitted use. The research brief points to the exact record supporting the claim, not merely a folder containing it.
Put the events in order.
Creation, transmission, receipt, review, approval and implementation are separate events. We retain original timestamps and note time zones, later edits and gaps. A document appearing in someone's file does not establish when they received it or what authority they held.
Contracts, scope schedules and issue registers help locate the boundary between who was present, who could decide and who actually signed off.
A higher budget is not the same as a paid loss.
A project budget, contract award, change order, invoice, payment and disputed claim answer different questions. We separate approved scope from contingency, owner costs, professional fees and unresolved amounts before comparing totals.
The useful crosswalk connects each change to its instruction, approval, valuation and payment record. It makes overlaps and missing support visible without counting the same amount twice.
Test another explanation.
Design, changed requirements, site conditions, installation, owner direction, maintenance and coordination can produce similar outcomes. For each explanation, we ask what mechanism is proposed, which records should exist, what contradicts it and what evidence would distinguish it from the alternatives.
New information can change the answer. Uncertainty belongs in the brief, alongside the finding and the next record needed to test it.
Request the record that would change the answer.
A useful records request identifies the project, custodian, record type, date range and decision it is meant to resolve. Executed agreements, change registers, signed instructions, approval records and final payment reconciliations often answer more than a broad request for every email.
We start with existing public records and documents you may lawfully share. External fees, confidential transfers and any technical access need an agreed scope and approval first.